How do I refresh a Power BI semantic model only after my ETL succeeds?
Turn off the scheduled refresh and start the refresh from whatever knows the load has finished: a Web activity at the end of the ADF or Fabric pipeline that calls the Power BI refresh API, a script, or an orchestrator task that depends on the load task with an On Success condition. A scheduled refresh only knows the time of day, so the night the load is late, the report refreshes on incomplete data.
Why scheduled refresh is the wrong trigger
A Power BI scheduled refresh starts at fixed times. If the load usually finishes at 02:15 and the refresh is at 03:00, the report is correct most nights and silently wrong on the nights the load runs late or fails. The fix is to make the refresh depend on the load, not on the clock.
Option 1: call the refresh API from the pipeline
The Power BI REST API starts a refresh with POST groups/{workspaceId}/datasets/{datasetId}/refreshes. Add a Web activity at the end of an ADF or Fabric pipeline, authenticated as a service principal or managed identity that is a member of the workspace.
Send a request body (for example a refresh type) to make it an enhanced refresh: it returns a refresh ID you can poll with GET .../refreshes/{refreshId}, so the pipeline can wait for the refresh result instead of firing and forgetting. Service-principal callers must also set notifyOption to NoNotification.
Option 2: an orchestrator dependency
With Polysync, the refresh is its own task on a Power BI platform:
- Connect the Power BI workspace; semantic models are discovered and imported.
- Create a Power BI Semantic Model Refresh task. Choose the refresh type (Full by default; Data Only, Calculate, Automatic or Clear Values are also available).
- Add a dependency: parent = the load task (ADF, Databricks, Fabric, Glue or any other), condition = On Success.
Polysync always sends an enhanced refresh, polls it to completion, shows failure detail from the refresh record, and can cancel a running refresh. If several models share a capacity, a contention profile on the Power BI platform keeps them from all refreshing at once.
Things that commonly go wrong
| Symptom | Likely cause |
|---|---|
| HTTP 401 or 403 | The identity is not a Member or Contributor on the workspace, or the tenant setting that allows service principals to use Power BI APIs is off |
| HTTP 409 | A refresh is already running for that model |
| Report still shows old data | The refresh succeeded but read from a source the load had not finished writing; check the dependency condition and the load's real completion |
Frequently asked questions
Can Power BI refresh automatically when new data arrives?
Not from a scheduled refresh. Something has to call the refresh API after the load. Pipelines, scripts and orchestrators can all do that.
Should I use a full refresh every time?
Not necessarily. For large models, Data Only or Calculate refreshes can be enough between full refreshes. Choose per task.
Can I refresh a dataflow the same way?
Yes. Power BI dataflows have their own refresh API, and Polysync has a separate Power BI Dataflow Refresh job type that can be chained the same way.
Microsoft, Azure, Azure Data Factory, Microsoft Fabric, Power BI, Logic Apps and Microsoft Marketplace are trademarks of the Microsoft group of companies. Apache Airflow and Airflow are trademarks of the Apache Software Foundation. Databricks and Lakeflow are trademarks of Databricks, Inc. All other product names are the property of their respective owners and are used for identification only. No endorsement by their owners is implied.